[Lint] Add `python/sglang` to ruff F401 checks and remove unused imports in files (#11685)

python/sglang/srt/models/gemma3n_mm.py

@@ -14,8 +14,7 @@ from transformers import (
 )
 from transformers.models.auto.modeling_auto import AutoModel
 
-from sglang.srt.layers.layernorm import RMSNorm
-from sglang.srt.layers.linear import ColumnParallelLinear, RowParallelLinear
+from sglang.srt.layers.linear import RowParallelLinear
 from sglang.srt.layers.logits_processor import LogitsProcessor
 from sglang.srt.layers.quantization.base_config import QuantizationConfig
 from sglang.srt.layers.vocab_parallel_embedding import VocabParallelEmbedding

python/sglang/srt/models/glm4_moe.py

@@ -44,10 +44,8 @@ from sglang.srt.layers.dp_attention import (
 )
 from sglang.srt.layers.layernorm import RMSNorm
 from sglang.srt.layers.linear import (
-    ColumnParallelLinear,
     MergedColumnParallelLinear,
     QKVParallelLinear,
-    ReplicatedLinear,
     RowParallelLinear,
 )
 from sglang.srt.layers.logits_processor import LogitsProcessor
@@ -78,16 +76,12 @@ from sglang.srt.utils import (
     BumpAllocator,
     LazyValue,
     add_prefix,
-    bind_or_assign,
     cpu_has_amx_support,
     get_bool_env_var,
     get_device_sm,
-    get_int_env_var,
     is_cpu,
     is_cuda,
-    is_flashinfer_available,
     is_hip,
-    is_non_idle_and_non_empty,
     log_info_on_rank0,
     use_intel_amx_backend,
 )

python/sglang/srt/models/gpt_oss.py

@@ -85,7 +85,7 @@ _is_sm100_supported = is_cuda() and is_sm100_supported()
 
 
 if _is_cuda:
-    from sgl_kernel import FusedSetKVBufferArg
+    from sgl_kernel import FusedSetKVBufferArg  # noqa: F401
 
 
 class GptOssConfig(PretrainedConfig):

python/sglang/srt/models/hunyuan.py

@@ -12,18 +12,14 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 """Inference-only HunYuan model compatible with HuggingFace weights."""
-import logging
 import re
-from dataclasses import dataclass
-from enum import Enum, auto
-from typing import Any, Dict, Iterable, List, Optional, Tuple, Union
+from typing import Any, Dict, Iterable, Optional, Tuple
 
 import torch
 from torch import nn
 from transformers import PretrainedConfig
 
 from sglang.srt.distributed import (
-    get_pp_group,
     get_tensor_model_parallel_rank,
     get_tensor_model_parallel_world_size,
     tensor_model_parallel_all_reduce,
@@ -46,7 +42,6 @@ from sglang.srt.layers.radix_attention import RadixAttention
 from sglang.srt.layers.rotary_embedding import get_rope
 from sglang.srt.layers.sampler import Sampler
 from sglang.srt.layers.vocab_parallel_embedding import (
-    DEFAULT_VOCAB_PADDING_SIZE,
     ParallelLMHead,
     VocabParallelEmbedding,
 )
@@ -56,7 +51,7 @@ from sglang.srt.model_loader.weight_utils import (
     kv_cache_scales_loader,
     maybe_remap_kv_scale_name,
 )
-from sglang.srt.utils import add_prefix, is_hip
+from sglang.srt.utils import is_hip
 
 expert_distribution_recorder = ExpertDistributionRecorder()
 

python/sglang/srt/models/interns1.py

@@ -5,7 +5,6 @@ from torch import nn
 from transformers import PretrainedConfig
 
 from sglang.srt.layers.attention import vision_utils
-from sglang.srt.layers.moe.ep_moe.layer import get_moe_impl_class
 from sglang.srt.layers.moe.fused_moe_triton.layer import FusedMoE
 from sglang.srt.layers.quantization.base_config import QuantizationConfig
 from sglang.srt.managers.mm_utils import (

python/sglang/srt/models/llama_eagle3.py

@@ -27,7 +27,7 @@ from transformers import LlamaConfig
 
 from sglang.srt.distributed import get_pp_group
 from sglang.srt.layers.layernorm import RMSNorm
-from sglang.srt.layers.linear import QKVParallelLinear, RowParallelLinear
+from sglang.srt.layers.linear import QKVParallelLinear
 from sglang.srt.layers.logits_processor import LogitsProcessor
 from sglang.srt.layers.quantization.base_config import QuantizationConfig
 from sglang.srt.layers.vocab_parallel_embedding import (

python/sglang/srt/models/longcat_flash.py

@@ -44,9 +44,7 @@ from sglang.srt.distributed import (
 )
 from sglang.srt.eplb.expert_distribution import get_global_expert_distribution_recorder
 from sglang.srt.eplb.expert_location import ModelConfigForExpertLocation
-from sglang.srt.eplb.expert_location_dispatch import ExpertLocationDispatchInfo
 from sglang.srt.layers.activation import SiluAndMul
-from sglang.srt.layers.amx_utils import PackWeightMethod
 from sglang.srt.layers.communicator import LayerCommunicator, LayerScatterModes
 from sglang.srt.layers.dp_attention import (
     get_attention_tp_rank,
@@ -87,20 +85,15 @@ from sglang.srt.models.deepseek_v2 import DeepseekV2AttentionMLA
 from sglang.srt.server_args import get_global_server_args
 from sglang.srt.utils import (
     BumpAllocator,
-    LazyValue,
     add_prefix,
     bind_or_assign,
     cpu_has_amx_support,
     get_bool_env_var,
     get_device_sm,
-    get_int_env_var,
     is_cpu,
     is_cuda,
-    is_flashinfer_available,
     is_hip,
-    is_non_idle_and_non_empty,
     is_npu,
-    is_sm100_supported,
 )
 
 _is_hip = is_hip()
@@ -113,13 +106,7 @@ _is_cpu = is_cpu()
 _device_sm = get_device_sm()
 
 if _is_cuda:
-    from sgl_kernel import (
-        awq_dequantize,
-        bmm_fp8,
-        dsv3_fused_a_gemm,
-        dsv3_router_gemm,
-        merge_state_v2,
-    )
+    from sgl_kernel import awq_dequantize
 elif _is_cpu and _is_cpu_amx_available:
     pass
 elif _is_hip:

python/sglang/srt/models/longcat_flash_nextn.py

@@ -32,14 +32,10 @@
 
 import concurrent.futures
 import logging
-import os
-from enum import IntEnum, auto
-from typing import Any, Dict, Iterable, Optional, Tuple, Union
+from typing import Iterable, Optional, Tuple
 
 import torch
-import torch.nn.functional as F
 from torch import nn
-from tqdm import tqdm
 
 from sglang.srt.configs import LongcatFlashConfig
 from sglang.srt.eplb.expert_distribution import get_global_expert_distribution_recorder
@@ -75,7 +71,6 @@ from sglang.srt.models.deepseek_v2 import DeepseekV2AttentionMLA
 from sglang.srt.models.longcat_flash import LongcatFlashForCausalLM, LongcatFlashMLP
 from sglang.srt.utils import (
     BumpAllocator,
-    LazyValue,
     add_prefix,
     bind_or_assign,
     cpu_has_amx_support,
@@ -97,13 +92,7 @@ _is_cpu = is_cpu()
 _device_sm = get_device_sm()
 
 if _is_cuda:
-    from sgl_kernel import (
-        awq_dequantize,
-        bmm_fp8,
-        dsv3_fused_a_gemm,
-        dsv3_router_gemm,
-        merge_state_v2,
-    )
+    from sgl_kernel import awq_dequantize
 elif _is_cpu and _is_cpu_amx_available:
     pass
 elif _is_hip:

python/sglang/srt/models/mimo.py

 # Adapted from qwen2.py
 
-from functools import partial
-from typing import Any, Dict, Iterable, Optional, Tuple
+from typing import Iterable, Optional, Tuple
 
 import torch
 from torch import nn
 
-from sglang.srt.distributed import (
-    get_tensor_model_parallel_rank,
-    get_tensor_model_parallel_world_size,
-    split_tensor_along_last_dim,
-    tensor_model_parallel_all_gather,
-)
-from sglang.srt.layers.layernorm import RMSNorm
-from sglang.srt.layers.linear import QKVParallelLinear, RowParallelLinear
 from sglang.srt.layers.logits_processor import LogitsProcessor
 from sglang.srt.layers.pooler import Pooler, PoolingType
 from sglang.srt.layers.quantization.base_config import QuantizationConfig
-from sglang.srt.layers.radix_attention import RadixAttention
-from sglang.srt.layers.rotary_embedding import get_rope
 from sglang.srt.layers.vocab_parallel_embedding import ParallelLMHead
 from sglang.srt.model_executor.forward_batch_info import ForwardBatch
 from sglang.srt.model_loader.weight_utils import default_weight_loader
-from sglang.srt.models.qwen2 import Qwen2DecoderLayer, Qwen2MLP, Qwen2Model
+from sglang.srt.models.qwen2 import Qwen2DecoderLayer, Qwen2Model
 from sglang.srt.utils import add_prefix
 
 MiMoConfig = None

python/sglang/srt/models/mimo_mtp.py

 # Adapted from https://github.com/vllm-project/vllm/pull/17433/files  and deepseek_nextn.py
 
-from functools import partial
-from typing import Any, Dict, Iterable, Optional, Tuple
+from typing import Iterable, Optional, Tuple
 
 import torch
 from torch import nn

python/sglang/srt/models/minicpmo.py

@@ -43,7 +43,6 @@ from sglang.srt.managers.mm_utils import (
     general_mm_embed_routine,
 )
 from sglang.srt.managers.schedule_batch import (
-    Modality,
     MultimodalDataItem,
     MultimodalInputs,
     flatten_nested_list,
@@ -59,8 +58,6 @@ from sglang.srt.utils import logger
 try:
     from transformers import LogitsWarper
     from vector_quantize_pytorch import GroupedResidualFSQ
-    from vocos import Vocos
-    from vocos.pretrained import instantiate_class
 
     _tts_deps = True
 except:

python/sglang/srt/models/mixtral.py

@@ -24,7 +24,6 @@ from torch import nn
 from transformers import MixtralConfig
 
 from sglang.srt.distributed import (
-    get_moe_expert_parallel_world_size,
     get_pp_group,
     get_tensor_model_parallel_world_size,
     tensor_model_parallel_all_reduce,

python/sglang/srt/models/opt.py

@@ -17,7 +17,6 @@ from collections.abc import Iterable
 from typing import Optional, Union
 
 import torch
-import torch.nn.functional as F
 from torch import nn
 from transformers import OPTConfig
 
@@ -26,10 +25,8 @@ from sglang.srt.distributed import (
     get_tensor_model_parallel_rank,
     get_tensor_model_parallel_world_size,
 )
-from sglang.srt.layers.activation import get_act_fn
 from sglang.srt.layers.linear import (
     ColumnParallelLinear,
-    MergedColumnParallelLinear,
     QKVParallelLinear,
     ReplicatedLinear,
     RowParallelLinear,
@@ -38,7 +35,7 @@ from sglang.srt.layers.logits_processor import LogitsProcessor, LogitsProcessorO
 from sglang.srt.layers.pooler import Pooler, PoolingType
 from sglang.srt.layers.quantization.base_config import QuantizationConfig
 from sglang.srt.layers.radix_attention import RadixAttention
-from sglang.srt.layers.utils import PPMissingLayer, get_layer_id
+from sglang.srt.layers.utils import get_layer_id
 from sglang.srt.layers.vocab_parallel_embedding import (
     ParallelLMHead,
     VocabParallelEmbedding,
@@ -47,7 +44,6 @@ from sglang.srt.model_executor.forward_batch_info import ForwardBatch, PPProxyTe
 from sglang.srt.model_loader.weight_utils import (
     default_weight_loader,
     kv_cache_scales_loader,
-    maybe_remap_kv_scale_name,
 )
 from sglang.srt.utils import add_prefix, make_layers
 

python/sglang/srt/models/phi.py

 # Adapted from https://github.com/vllm-project/vllm/blob/main/vllm/model_executor/models/phi.py
-from typing import Iterable, Optional, Union
+from typing import Iterable, Optional
 
 import torch
 from torch import nn

python/sglang/srt/models/phi4mm.py

@@ -24,7 +24,7 @@ from typing import List, Optional, Tuple
 import numpy as np
 import torch
 from torch import nn
-from transformers import PretrainedConfig, SiglipVisionConfig
+from transformers import PretrainedConfig
 
 from sglang.srt.layers.quantization import QuantizationConfig
 from sglang.srt.managers.mm_utils import (

python/sglang/srt/models/phimoe.py

@@ -18,7 +18,6 @@ from sglang.srt.layers.pooler import Pooler, PoolingType
 from sglang.srt.layers.quantization.base_config import QuantizationConfig
 from sglang.srt.layers.radix_attention import RadixAttention
 from sglang.srt.layers.rotary_embedding import get_rope
-from sglang.srt.layers.utils import PPMissingLayer
 from sglang.srt.layers.vocab_parallel_embedding import (
     DEFAULT_VOCAB_PADDING_SIZE,
     ParallelLMHead,

python/sglang/srt/models/pixtral.py

@@ -16,13 +16,10 @@
 Using mistral-community/pixtral-12b as reference.
 """
 
-import logging
-import math
 from typing import Iterable, List, Optional, Set, Tuple, Union
 
 import torch
 import torch.nn as nn
-import torch.nn.functional as F
 from transformers import PixtralVisionConfig, PretrainedConfig
 from transformers.models.pixtral.modeling_pixtral import PixtralRotaryEmbedding
 from transformers.models.pixtral.modeling_pixtral import (

python/sglang/srt/models/qwen.py

@@ -15,7 +15,6 @@
 # Adapted from
 # https://github.com/vllm-project/vllm/blob/c7f2cf2b7f67bce5842fedfdba508440fe257375/vllm/model_executor/models/qwen.py#L1
 
-import time
 from typing import Any, Dict, Iterable, Optional, Tuple
 
 import torch

python/sglang/srt/models/qwen2_audio.py

@@ -23,30 +23,18 @@
 # limitations under the License.
 """Inference-only Qwen2-Audio model compatible with HuggingFace weights."""
 import logging
-import math
-from functools import lru_cache, partial
-from typing import Any, Iterable, List, Optional, Tuple, Type, TypedDict
+from typing import Any, Iterable, List, Optional, Tuple
 
 import torch
 import torch.nn as nn
-import torch.nn.functional as F
-from einops import rearrange
-from transformers import AutoTokenizer, Qwen2AudioEncoderConfig, Qwen2Config
-from transformers.activations import ACT2FN
+from transformers import Qwen2AudioEncoderConfig, Qwen2Config
 from transformers.models.qwen2_audio.configuration_qwen2_audio import Qwen2AudioConfig
 from transformers.models.qwen2_audio.modeling_qwen2_audio import (
     Qwen2AudioEncoder,
     Qwen2AudioMultiModalProjector,
 )
 
-from sglang.srt.layers.activation import QuickGELU
-from sglang.srt.layers.attention.vision import VisionAttention
-from sglang.srt.layers.linear import ColumnParallelLinear, RowParallelLinear
-from sglang.srt.layers.logits_processor import LogitsProcessor
-from sglang.srt.layers.pooler import Pooler, PoolingType
 from sglang.srt.layers.quantization.base_config import QuantizationConfig
-from sglang.srt.layers.utils import get_layer_id
-from sglang.srt.layers.vocab_parallel_embedding import ParallelLMHead
 from sglang.srt.managers.mm_utils import (
     MultiModalityDataPaddingPatternMultimodalTokens,
     general_mm_embed_routine,
@@ -60,7 +48,6 @@ from sglang.srt.model_executor.forward_batch_info import ForwardBatch
 from sglang.srt.model_loader.weight_utils import default_weight_loader
 from sglang.srt.models.qwen2 import Qwen2ForCausalLM
 from sglang.srt.utils import add_prefix
-from sglang.srt.utils.hf_transformers_utils import get_processor
 
 logger = logging.getLogger(__name__)
 

python/sglang/srt/models/qwen3_next.py

 import enum
 import logging
-from typing import Any, Dict, Iterable, Optional, Set, Tuple
+from typing import Any, Iterable, Optional, Set, Tuple
 
 import torch
-import torch.nn.functional as F
 from torch import nn
 
 from sglang.srt.configs.qwen3_next import Qwen3NextConfig
-from sglang.srt.distributed import (
-    divide,
-    get_pp_group,
-    get_tensor_model_parallel_rank,
-    get_tensor_model_parallel_world_size,
-)
+from sglang.srt.distributed import divide, get_pp_group
 from sglang.srt.eplb.expert_distribution import get_global_expert_distribution_recorder
 from sglang.srt.eplb.expert_location import ModelConfigForExpertLocation
 from sglang.srt.layers.attention.fla.layernorm_gated import RMSNorm as RMSNormGated
@@ -23,10 +17,9 @@ from sglang.srt.layers.dp_attention import (
     get_attention_tp_size,
     is_dp_attention_enabled,
 )
-from sglang.srt.layers.layernorm import GemmaRMSNorm, RMSNorm
+from sglang.srt.layers.layernorm import GemmaRMSNorm
 from sglang.srt.layers.linear import (
     ColumnParallelLinear,
-    MergedColumnParallelLinear,
     QKVParallelLinear,
     RowParallelLinear,
 )